C51 uses the 32-bit and 16-bit arithmetic operations of the 80C517 and 80C537 to improve performance of a number of math-intensive operations. C language programs execute considerably faster when using either of these CPUs. The following tables show execution times for various arithmetic operations and compares the performance of the standard 8051 to that of the 80C517 CPU.
| 16-bit Binary Integer Operations |
Execution Time in CPU Cycles |
|||
| Operation/CPU |
Library Routine |
Min. |
Typ. |
Max. |
| Signed/unsigned multiplication |
||||
| 8051 |
IMUL |
31 |
||
| 80517 |
intrinsic |
16 |
||
| Unsigned division |
||||
| 8051 |
UIDIV |
18 |
100 |
190 |
| 80517 |
UIDIV517 |
24 |
||
| Signed division |
||||
| 8051 |
SIDIV |
38 |
120 |
210 |
| 80517 |
SIDIV517 |
44 |
||
| 32-bit Binary Integer Operations |
Execution Time in CPU Cycles |
|||
| Operation/CPU |
Library Routine |
Min. |
Typ. |
Max. |
| Signed/unsigned multiplication |
||||
| 8051 |
LMUL |
128 |
||
| 80517 |
LMUL517 |
74 |
||
| Unsigned division |
||||
| 8051 |
ULDIV |
1000 |
1100 |
1500 |
| 80517 |
ULDIV517 |
38 |
80 |
126 |
| Signed division |
||||
| 8051 |
SLDIV |
1030 |
1130 |
1530 |
| 80517 |
SLDIV517 |
68 |
90 |
156 |
| Left shift |
||||
| 8051 |
LSHL |
7 |
7+(15×cnt) |
472 |
| 80517 |
LSHL517 |
7 |
30 |
31 |
| Unsigned right shift |
||||
| 8051 |
ULSHR |
7 |
7+(15×cnt) |
472 |
| 80517 |
ULSHR517 |
7 |
31 |
32 |
| Signed right shift |
||||
| 8051 |
SLSHR |
7 |
7+(15×cnt) |
472 |
| 80517 |
SLSHR517 |
7 |
31 |
32 |
| Floating-point Operations |
Execution Time in CPU Cycles |
|||
| Operation/CPU |
Library Routine |
Min. |
Typ. |
Max. |
| Addition |
||||
| 8051 |
FPADD |
100 |
220 |
285 |
| 80517 |
FPADD517 |
100 |
190 |
205 |
| Subtraction |
||||
| 8051 |
FPSUB |
100 |
220 |
287 |
| 80517 |
FPSUB517 |
100 |
190 |
207 |
| Multiplication |
||||
| 8051 |
FPMUL |
352 |
375 |
384 |
| 80517 |
FPMUL517 |
200 |
215 |
230 |
| Division |
||||
| 8051 |
FPDIV |
1020 |
1150 |
1210 |
| 80517 |
FPDIV517 |
250 |
285 |
350 |
| Comparison |
||||
| 8051 |
FPCMP |
100 |
||
| 80517 |
FPCMP517 |
100 |
||
| Square root |
||||
| 8051 |
SQRT |
40 |
2560 |
2860 |
| 80517 |
SQRT517 |
40 |
1150 |
1170 |
| Sine |
||||
| 8051 |
SIN |
4770 |
||
| 80517 |
SIN517 |
3500 |
||
| Cosine |
||||
| 8051 |
COS |
5300 |
||
| 80517 |
COS517 |
3700 |
||
| Tangent |
||||
| 8051 |
TAN |
2370 |
9570 |
|
| 80517 |
TAN517 |
2370 |
6170 |
|
| Arcsine |
||||
| 8051 |
ASIN |
930 |
9730 |
10060 |
| 80517 |
ASIN517 |
740 |
5280 |
5370 |
| Arccosine |
||||
| 8051 |
ACOS |
2030 |
10250 |
11470 |
| 80517 |
ACOS517 |
1850 |
5800 |
5900 |
| Arctangent |
||||
| 8051 |
ATAN |
1125 |
6700 |
|
| 80517 |
ATAN517 |
1125 |
3760 |
|
| Exponential |
||||
| 8051 |
EXP |
2940 |
7680 |
7870 |
| 80517 |
EXP517 |
2840 |
5380 |
5530 |
| Logarithm |
||||
| 8051 |
LOG |
45 |
5730 |
6050 |
| 80517 |
LOG517 |
45 |
3550 |
3600 |
| ASCII to floating-point conversion |
||||
| 8051 |
FPATOF |
800 |
3000 |
16380 |
| 80517 |
FPATOF517 |
680 |
2050 |
9930 |
NOTE
The execution times specified in this table do not take into
consideration access times for variables or stack operations. The
actual processing times may consume up to 100 additional cycles
depending on the accessed address space and stack load.
When using the arithmetic features of the 80C517 and 80C537, note that the operations involving the arithmetic processor are exclusive and are not interrupt-capable. For this reason, do not use the arithmetic extensions in both the main program and in an interrupt service routine. By doing the following, you can guarantee that only one thread of execution uses the arithmetic processor:
Use the MOD517 directive to compile functions which are guaranteed to be executed only in the main program or functions used by one interrupt service routine, but not both.
Compile all remaining functions with the MOD517(NOAU) directive.